<div class="MsoNormal">An straightforward answer to this question may be just experience. One can monitor the convergence procedures during two given approaches to realize why it is recommended to include gradually the spin-orbit coupling (SOC) to a well previously converged SCF in the absence of SOC. This answer can satisfy everyone in the case of observing considerable improvement on the convergence procedure (as discussed before by <tt>Lyudmila</tt>).</div> <div class="MsoNormal">However, to be more curios one can back to the time-independent perturbation theory. In this theory the perturbed Hamiltonians are diagonalized in the space of non-perturbed eigenstates.</div> <div class="MsoNormal">If we take SOC term as perturbed Hamiltonians, then performing an SCF without SOC generates the scalar-relativistic eigenstates, case.vector. Now including SOC can be interpreted as a diagonalization of the SOC term in the generated space of
case.vectorup(dn) to generate new eigenstates of case.vectorso. This will reduce the oscillation and save the time as well. In the presence of SOC the up and down spins are not longer independent leading to a non-diagonal matrix and thereby we encounter a doubled matrix which should be <span style=""> </span>diagonalized from scratch. </div> <br><br><b><i>"Atta-fynn, Raymond" <attafynn@uta.edu></i></b> wrote:<blockquote class="replbq" style="border-left: 2px solid rgb(16, 16, 255); margin-left: 5px; padding-left: 5px;"> <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1"> <div id="idOWAReplyText53834" dir="ltr"> <div dir="ltr"><font color="#000000" face="Arial" size="2">Dear WIEN users,</font></div> <div dir="ltr"><font face="Arial" size="2"></font> </div> <div dir="ltr"><font face="Arial" size="2">It is clearly stated in the manual that to do Spin-orbit (SO) computation, </font></div> <div dir="ltr"><font face="Arial"
size="2">one has to run a regular scf calculation, save the results, initialize SO, and run </font></div> <div dir="ltr"><font face="Arial" size="2">the SO calculation, that is</font></div> <div dir="ltr"><font face="Arial" size="2"></font> </div> <div dir="ltr"><font face="Arial" size="2">(1) runsp_lapw -i 100 -ec 0.00001</font></div> <div dir="ltr"><font face="Arial" size="2">(2) save results</font></div> <div dir="ltr"><font face="Arial" size="2">(3) initso</font></div> <div dir="ltr"><font face="Arial" size="2">(4) runsp_lapw -so -i 100 -ec 0.00001</font></div> <div dir="ltr"> </div> <div dir="ltr"><font face="Arial" size="2">My question is why run non-SO first followed by SO. </font></div> <div dir="ltr"><font face="Arial" size="2">What is wrong with just initializing SO and going straight ahead </font></div> <div dir="ltr"><font face="Arial" size="2">with the SO calculation (that is just execute steps
3 and 4 and skip step 1).</font></div> <div dir="ltr"> </div> <div dir="ltr"><font face="Arial" size="2">Thanks</font></div> <div dir="ltr"> </div> <div dir="ltr"><font face="Arial" size="2">R. Atta-Fynn</font> </div> <div dir="ltr"><font color="#000000" face="Arial" size="2"></font> </div></div> <div id="idSignature53438" dir="ltr"> <div><font color="#000000" face="Arial" size="2"></font> </div></div>_______________________________________________<br>Wien mailing list<br>Wien@zeus.theochem.tuwien.ac.at<br>http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien<br></blockquote><br><BR><BR>Sincerely yours,<br>S. Jalali.<br>/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/<br>Saeid Jalali Asadabadi,<br>Department of Physics, Faculty of Science,<br>University of Isfahan (UI), Hezar Gerib Avenue,<br>81744 Isfahan, Iran.<br>Phones:<br>Dep. of Phys. :+98-0311-793 2435<br>Office :+98-0311-793 2430<br>Fax No. :+98-0311-793
2409<br>E-mail :s_jalali_a@yahoo.com<br>www :http://www.ui.ac.ir<br>/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/<p> 
<hr size=1>Groups are talking. We´re listening. Check out the <a href="http://pa.yahoo.com/*http://us.rd.yahoo.com/evt=41144/*http://groups.yahoo.com/local/newemail.html">handy changes to Yahoo! Groups.</a>